This study proposes to develop objective analytic methods to help resolve a problem that claims the lives of over 16,000 Americans per year. Alcohol-impaired driving is difficult to detect and accordingly the annual 1.5 million DUI convictions in the US represent less than a tenth of 1 percent of all episodes of impaired driving. The Healthy People 2010 target to reduce alcohol-related motor vehicle fatalities by 31 percent will require better methods for identifying high-risk drivers, and more attainable safety interventions. This study proposes to recruit 500 multiple DUI offenders who will serve as test cases for objective alcohol-risk profiles. The DUI offenders to be studied will have been assigned to use an alcohol ignition interlock device as a step on the path to license reinstatement. The interlock requires a low BAC or alcohol-free breath test before an engine can be started. In addition to blocking impaired driving, the interlock record of BAC tests _>.04 percent is a dependable advance indicator of future (post-interlock) repeat DUI. The study will evaluate two categories of objective markers: behavioral markers based on the alcohol interlock record of BAC tests, and blood-borne biomarkers of high alcohol consumption (CDT, GGT, MCV). The blood-borne biomarkers are often found to be high specificity indicators of heavy alcohol consumption. These markers together with driver records and validated self-report assessment instruments (TLFB, AUDIT, DIS-AM, TRI, DrInC) will be used to develop predictive profiles for future DUI that can serve as a basis for prospective sentencing guidelines to lower public risk exposure. Subjects will provide blood samples and fill in assessments when the interlock is installed and when it is removed. Two types of endpoint will be predicted from multiple indicators: future repeat DUI (AIM 1) and failed interlock BAC tests (AIM 2). A third AIM will determine if repeated measures, at interlock installation and removal, improve decision resolution. The project will use segmentation models, survival analysis methods and signal detection approaches to evaluate diagnostic tools. With objective screening methods, prescriptive approaches can be derived that support practical criterion-based sanctions and license decisions. In addition to scientific communication through usual channels, this project will create risk profile guidelines to aid motor vehicle authorities and the courts for using indicator evidence proactively.